Multi-Period Performance Attribution

Here is what I found interesting part of it, and would like to share with you. Perhaps this has been pointed out, then ignore what I said. I notice that the two formulas on Schweser Book 5(Page 150) look different from those in other places. 1, SSE(security selection effect) uses w(p,j), which is consistent with Global Attribution formula. [Micro Attribution on page 94 uses w(b,j)] 2, MAE(market allocation effect) uses -Rb, which is consistent with Micro Attribution formula. [Global Attribution on page 142 doesn’t use Rb] Schwser does point out this in Book 5, but I didn’t understand it until I read CFAI book. All these things make Multi-period performance attribution more complicated. I also think the following formula interesting [Rp1=Rb1+Ra1, so, Rp1 depends on Ra1]. R(A,2) = Ra1*(1+Rb2) + Ra2*(1+Rp1) Now I understand a little bit why it’s doing that…it’s trying to make the two period’s active return additive…But, because of this Ra1*Ra2, the addition and compounding still won’t work.

the point is that the return’s are not additive. looking at the R(A,2) = Ra1*(1+Rb2) + Ra2*(1+Rp1) formula, you can see the active return in period 2, is partly because of the benchmark return on the active gain from period 1, and partly from the active return in period 2 on the total return from the 1st period. in period 1 you can split into benchmark return and active return, each of these components will grow at the benchmark, and also at the active rate in period 2. so there’s 3 components to the active return over both periods: active return p1, active return (p2) on benchmark return (p1) and benchmark return (p2) on active return p1

A couple spreadsheets I built on 47.c and 47.e helped me grasp this material:

gjertsen Wrote: ------------------------------------------------------- > A couple spreadsheets I built on 47.c and 47.e > helped me grasp this material: > > > tfolio-return-attribution/ Thank you very much for your both presentations!

Hi Deriv108, I noticed the same things. 2) Using or not -Rb gives in both case the same total sum result. By not substracting Rb, it simplifies the formula, but doesn’t allow you to find the attribution for each sectors, but only the total sum. Schweser point this out but doesn’t go far in explanation. 1)I still don’t catch why it’s w(p,j) and not w(b,j) in the global attribution. Bern